This article is more than 1 year old

NetApp product refresh time

Your three years is up

The NetApp part of the twitterverse has been full of scarcely suppressed excitement about WAFL and solid state drives (SSDs). Could NetApp be about to open its SSD kimono at SNW Dallas on October 11th?

NetApp has been sending out pro-SSD messages and then not fulfilling them for a couple of years. In November 2008 it introduced a mid-range FAS3160 array and stated: "NetApp also unveiled today that the new FAS3160, V3160, and other FAS and V-Series storage systems will be enterprise flash ready. As part of its innovative strategy, NetApp will combine flash memory with its unified storage architecture to meet the performance needs of both leading-edge and mainstream IT professionals. Unlike other storage vendors, NetApp will give customers practical options by using flash memory both as persistent storage and also for caching purposes."

Then, in August 2009, NetApp exec Jay Kidd said its storage arrays would get flash in the drive enclosures by the end of the year, and use a SAS interface.

Huh; here we are, almost two years after the FAS3160 announcement, a year after Kidd's promise, and NetApp has concentrated on its Flash Cache technology, adding flash to its controllers, leaving its drive enclosures a flash-free zone. It's getting overdue we might say, with NetApp virtually the only mainstream storage array supplier without flash in its drive shelves.

El Reg is hearing noises that NetApp has a hardware refresh or two coming up. WAFL - NetApp's Write Anywhere File Layout - is the underpinning of NetApp's O/S, Data ONTAP, currently in v8.0 form. It runs on three basic product ranges. The top end FAS6080 and 6040 products were introduced in December, 2007; NetApp tends to have a 3-year hardware refresh cycle by the way.

The mid-range FAS3160 was first revealed in November 2008 with the 3140 and 3170 appearing six months before it. The FAS2000 is older with the 2030 and 2050 coming around the corner in September 2007, and the 2040 in September 2009.

We might consider that NetApp could refresh both the FAS6000 and FAS2000 product groups based on a 3-year refresh cycle.

We note that Alex MacDonald from NetApp's Office of the CTO is presenting on parallel NFS (NFS 4.1) at SNW in Dallas next week.

If we were to erect an announcement castle on these grains of sand we might say NetApp could announce refreshed FAS6000 and FAS2000 hardware - FAS6100 and FAS2100 anyone? - with ONTAP 8.1 supporting parallel NFS (NFS 4.1) and SSDs.

Our take is that both single level cell (SLC) and multi-level cell (MLC) flash will be used, based on NetApp twitterers. Other comment from the denizens of Sunnyvale says WAFL doesn't do in-place updates and can circumvent in some way the copy-write-erase cycle of flash write updates; in fact WAFL might even eliminate the well-known MLC flash wear problem.

One person says that NetApp is already delivering SLC flash and is working on MLC support. Some kind of auto-tiering is also going to be offered.

Micron has announced its enterprise MLC or eMLC flash, which is being used by Pliant and Viking Modular. Perhaps NetApp is planting its SSD flag on Micron's eMLC turf?

The 6000 and 3100 product groups use a FCAL backend link between the controllers and the disk drives, and support 4Gbit/s Fibre Channel host connectivity. The FAS2000s support Fibre Channel, SAS and SATA drives while the 6000 supports just Fibre Channel and SATA ones. Perhaps they will get a SAS backend link between controllers and drives and full SAS drive support.

While NetApp is about it, there could be added support for 8Gbits Fibre Channel host connectivity.

Striping problem

We also hear from someone who is quite intimately associated with the situation, that ONTAP 8 doesn't support striping across controllers. Customers with the FAS350, which is no longer available for purchase, and which runs ONTAP GX, will have to migrate to a FAS3040 or 3070, and there ONTAP 8 doesn't support the 3050's striping across controllers.

This view is apparently contradicted by NetApp's marketing literature, which states:

"Data ONTAP 8.0 Cluster-Mode systems use clustered file-system technology to provide maximum I/O throughput and remove the bottlenecks that impact production. You can stripe volumes across any or all of the storage controllers and disks in the system."

On having this pointed out, our source double checked inside NetApp's engineering organisation on the issue. The response was that: "Stripe across heads is gone and it is not coming back because it did not work with SnapMirror, snapshots, etc." ®

More about

TIP US OFF

Send us news


Other stories you might like